Re: Issues Outstanding for Point In Time Recovery (PITR)

Zeugswetter Andreas SB SD wrote:
> 
> > OK, so you do a tar backup of a file.  While you are doing the tar,
> > certain 8k blocks are being modified in the file.  There is no way to
> > know what blocks are modified as you are doing the tar, and in fact you
> > could read partial page writes during the tar.
> 
> No, I think all OS's (Unix and NT at least) guard against this, as long as 
> the whole 8k block is written in one call. It is only the physical layer (disk) 
> that is prone to partial writes.

Yes, good point.  The kernel will present a unified view of the 8k
block.  Of course, there are still cases where 8k blocks are being
changed in front/behind in the tarred file.  Will WAL allow us to
re-synchronize that file even if part of it has pages from an earlier in
time than other pages.  Uh, I think so.  So maybe we don't need the
pre-write images in WAL after all.  Can we replay the WAL when some
pages in the restored file _have_ the WAL changes and some don't?  Maybe
the LSN on the pages helps with this.

--  Bruce Momjian                        |  http://candle.pha.pa.us pgman@candle.pha.pa.us               |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026